Moderating Large Instances
To limit the effects of the recent spam wave from mastodon.social we temporarily limited their instance (that limit has since been removed). As a reminder, limiting an instance has the following effects:
If you are following a user at the limited instance there is no difference (you can see their profile, interact with them, etc.)
You will not see notification of mentions from users of the limited instance that you are not following
If you visit a user on the limited instances profile you will see a warning saying that your admin has limited them, but you can click through to see their content and follow them if you want them to be able to message you
Unfortunately the tools Mastodon gives us to deal with spam waves are really this, or a full suspension (no one can interact with that instance).
Unfortunately, Mastodon.social is a huge instance that is promoted to new users of the fediverse by the "official" Mastodon clients and services such as joinmastodon.org and by people in the Mastodon company. This gives it a "too big to fail" paradox where it's hard to moderate because many new users of the fediverse will find themselves "banned" from their friends servers, causing confusion and the common "the fediverse is too complicated" complaint, but also makes it a prime target for spam.
Large, general-purpose instances like mastodon.social have historically been a weak point in the network that is under-moderated, so this discussion is a place to brainstorm and come up with ideas. Should we limit mastodon.social permanently? Is this trying to solve a problem that doesn't exist? Should we take even more aggressive action against mastodon.social to try and limit the spread of too-big-to-fail instances? The Community Working Groups ops team would love to get your feedback on how we handled this specific issue, and how we should handle similar issues going forward. Thanks!
Erik Moeller · Fri 5 May 2023 9:39PM
Thanks for kicking off this discussion. I think the approach you took to limit mastodon.social temporarily was the correct one. Thanks to all involved in making and implementing that decision.
On the bigger questions you raise, while I think many of us would prefer a more decentralized fediverse, I don't think that preference alone should cause us to cut off or constrain people's ability to communicate with folks from any one instance.
This may be a long shot given our limited resources, but perhaps it's possible to define quantitative criteria (number of spam message, time it takes to suspend users who violate community norms, etc.) that could help us in long-term decision making?